In [ ]:
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import os
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename))
break
/kaggle/input/testssss/testdata.csv /kaggle/input/testssss/test_crop_image/wheat-field03.jpg /kaggle/input/agriculture-crop-images/Crop_details.csv /kaggle/input/agriculture-crop-images/kag2/rice/rice005arot.jpeg /kaggle/input/agriculture-crop-images/kag2/maize/maize034a.jpeg /kaggle/input/agriculture-crop-images/kag2/jute/jute014a.jpeg /kaggle/input/agriculture-crop-images/kag2/wheat/wheat016arot.jpeg /kaggle/input/agriculture-crop-images/kag2/sugarcane/sugarcane007ahs.jpeg /kaggle/input/agriculture-crop-images/crop_images/rice/rice017a.jpeg /kaggle/input/agriculture-crop-images/crop_images/maize/maize034a.jpeg /kaggle/input/agriculture-crop-images/crop_images/jute/jute014a.jpeg /kaggle/input/agriculture-crop-images/crop_images/wheat/wheat036a.jpeg /kaggle/input/agriculture-crop-images/crop_images/sugarcane/sugarcane0001a.jpeg
In [2]:
%matplotlib inline
import warnings
warnings.filterwarnings('ignore')
from sklearn.preprocessing import OneHotEncoder
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
import seaborn as sns
import tensorflow as tf
import keras
from keras.layers import Conv2D
from keras.models import Sequential
from keras.layers import MaxPool2D
from keras.layers import Flatten
from keras.layers import Dense
from keras.models import Model
from keras.preprocessing.image import ImageDataGenerator
from sklearn.metrics import confusion_matrix
from sklearn.metrics import accuracy_score
import cv2
import re
import random
random.seed(0)
np.random.seed(0)
Plotting some images¶
In [3]:
wheat = plt.imread("/kaggle/input/agriculture-crop-images/kag2/wheat/wheat0004a.jpeg")
jute = plt.imread("/kaggle/input/agriculture-crop-images/kag2/jute/jute005a.jpeg")
cane = plt.imread("/kaggle/input/agriculture-crop-images/kag2/sugarcane/sugarcane0010arot.jpeg")
rice = plt.imread("/kaggle/input/agriculture-crop-images/kag2/rice/rice032ahs.jpeg")
maize = plt.imread("/kaggle/input/agriculture-crop-images/kag2/maize/maize003a.jpeg")
plt.figure(figsize=(20,3))
plt.subplot(1,5,1)
plt.imshow(jute)
plt.title("jute")
plt.subplot(1,5,2)
plt.imshow(maize)
plt.title("maize")
plt.subplot(1,5,3)
plt.imshow(rice)
plt.title("rice")
plt.subplot(1,5,4)
plt.imshow(cane)
plt.title("sugarcane")
plt.subplot(1,5,5)
plt.imshow(wheat)
plt.title("wheat")
Out[3]:
Text(0.5, 1.0, 'wheat')
In [4]:
jutepath = "../input/agriculture-crop-images/kag2/jute"
maizepath = "../input/agriculture-crop-images/kag2/maize"
ricepath = "../input/agriculture-crop-images/kag2/rice"
sugarcanepath = "../input/agriculture-crop-images/kag2/sugarcane"
wheatpath = "../input/agriculture-crop-images/kag2/wheat"
jutefilename = os.listdir(jutepath)
maizefilename = os.listdir(maizepath)
ricefilename = os.listdir(ricepath)
sugarcanefilename = os.listdir(sugarcanepath)
wheatfilename = os.listdir(wheatpath)
X= []
In [5]:
for fname in jutefilename:
X.append([os.path.join(jutepath,fname),0])
for fname in maizefilename:
X.append([os.path.join(maizepath,fname),1])
for fname in ricefilename:
X.append([os.path.join(ricepath,fname),2])
for fname in sugarcanefilename:
X.append([os.path.join(sugarcanepath,fname),3])
for fname in wheatfilename:
X.append([os.path.join(wheatpath,fname),4])
X = pd.DataFrame(X,columns = ['path','labels'])
In [6]:
X.head()
Out[6]:
| path | labels | |
|---|---|---|
| 0 | ../input/agriculture-crop-images/kag2/jute/jut... | 0 |
| 1 | ../input/agriculture-crop-images/kag2/jute/jut... | 0 |
| 2 | ../input/agriculture-crop-images/kag2/jute/jut... | 0 |
| 3 | ../input/agriculture-crop-images/kag2/jute/jut... | 0 |
| 4 | ../input/agriculture-crop-images/kag2/jute/jut... | 0 |
In [7]:
ohencoder = OneHotEncoder(handle_unknown='ignore',sparse=False)
ohlabel = pd.DataFrame(ohencoder.fit_transform(X[['labels']]),dtype = 'float64',columns = ['label0','label1','label2','label3','label4'])
label_X = X.copy()
X = pd.concat([X,ohlabel],axis = 1)
new_X = X.drop(['labels'],axis = 1)
In [8]:
train,test = train_test_split(new_X,test_size=0.2,random_state=32,shuffle = True)
In [9]:
X_train = train['path'].values
y_train = train.drop(['path'],axis=1).values
X_test = test['path'].values
y_test = test.drop(['path'],axis=1).values
In [10]:
def deep_pipeline(data):
flat = []
for i in data:
img = plt.imread(i)
img = img/255.
flat.append(img)
flat = np.array(flat)
flat = flat.reshape(-1,224,224,3)
return flat
In [11]:
dx_train = deep_pipeline(X_train)
dx_test = deep_pipeline(X_test)
In [12]:
keras.backend.clear_session()
vgg = keras.applications.VGG19(input_shape=(224,224,3),include_top=False,weights = 'imagenet',pooling='avg')
vgg.trainable = False
vggmodel = keras.Sequential([vgg
,Dense(1000,activation='tanh'),Dense(1000,activation='tanh'),Dense(1000,activation='tanh'),Dense(5,activation='softmax')])
vggmodel.compile(optimizer = 'adam',loss = 'categorical_crossentropy',metrics=['accuracy'])
vggmodel.summary()
Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg19/vgg19_weights_tf_dim_ordering_tf_kernels_notop.h5 80142336/80134624 [==============================] - 1s 0us/step Model: "sequential" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= vgg19 (Model) (None, 512) 20024384 _________________________________________________________________ dense (Dense) (None, 1000) 513000 _________________________________________________________________ dense_1 (Dense) (None, 1000) 1001000 _________________________________________________________________ dense_2 (Dense) (None, 1000) 1001000 _________________________________________________________________ dense_3 (Dense) (None, 5) 5005 ================================================================= Total params: 22,544,389 Trainable params: 2,520,005 Non-trainable params: 20,024,384 _________________________________________________________________
In [13]:
hist = vggmodel.fit(dx_train,y_train,epochs=50,validation_split=0.3,batch_size=16)
Epoch 1/50 28/28 [==============================] - 3s 109ms/step - loss: 3.0200 - accuracy: 0.2780 - val_loss: 1.3661 - val_accuracy: 0.4792 Epoch 2/50 28/28 [==============================] - 2s 76ms/step - loss: 1.2730 - accuracy: 0.5157 - val_loss: 1.2540 - val_accuracy: 0.5417 Epoch 3/50 28/28 [==============================] - 2s 77ms/step - loss: 1.0431 - accuracy: 0.5673 - val_loss: 1.2434 - val_accuracy: 0.5156 Epoch 4/50 28/28 [==============================] - 2s 76ms/step - loss: 0.8738 - accuracy: 0.6839 - val_loss: 1.1318 - val_accuracy: 0.5260 Epoch 5/50 28/28 [==============================] - 2s 77ms/step - loss: 0.7871 - accuracy: 0.6996 - val_loss: 0.6435 - val_accuracy: 0.7500 Epoch 6/50 28/28 [==============================] - 2s 76ms/step - loss: 0.7050 - accuracy: 0.7220 - val_loss: 0.6128 - val_accuracy: 0.7552 Epoch 7/50 28/28 [==============================] - 2s 76ms/step - loss: 0.6688 - accuracy: 0.7803 - val_loss: 0.8669 - val_accuracy: 0.6406 Epoch 8/50 28/28 [==============================] - 2s 77ms/step - loss: 0.4596 - accuracy: 0.8206 - val_loss: 0.5200 - val_accuracy: 0.7760 Epoch 9/50 28/28 [==============================] - 2s 76ms/step - loss: 0.4923 - accuracy: 0.8117 - val_loss: 0.7053 - val_accuracy: 0.7604 Epoch 10/50 28/28 [==============================] - 2s 75ms/step - loss: 0.3556 - accuracy: 0.8632 - val_loss: 0.4693 - val_accuracy: 0.8333 Epoch 11/50 28/28 [==============================] - 2s 76ms/step - loss: 0.5160 - accuracy: 0.8318 - val_loss: 0.9860 - val_accuracy: 0.6875 Epoch 12/50 28/28 [==============================] - 2s 78ms/step - loss: 0.4851 - accuracy: 0.8251 - val_loss: 0.8337 - val_accuracy: 0.7188 Epoch 13/50 28/28 [==============================] - 2s 77ms/step - loss: 0.4954 - accuracy: 0.8139 - val_loss: 1.0346 - val_accuracy: 0.7135 Epoch 14/50 28/28 [==============================] - 2s 77ms/step - loss: 0.3024 - accuracy: 0.9058 - val_loss: 0.4834 - val_accuracy: 0.8333 Epoch 15/50 28/28 [==============================] - 2s 80ms/step - loss: 0.2474 - accuracy: 0.9036 - val_loss: 0.4063 - val_accuracy: 0.8854 Epoch 16/50 28/28 [==============================] - 2s 79ms/step - loss: 0.1173 - accuracy: 0.9619 - val_loss: 0.4649 - val_accuracy: 0.8802 Epoch 17/50 28/28 [==============================] - 2s 76ms/step - loss: 0.1102 - accuracy: 0.9529 - val_loss: 0.4781 - val_accuracy: 0.8854 Epoch 18/50 28/28 [==============================] - 2s 78ms/step - loss: 0.1052 - accuracy: 0.9552 - val_loss: 0.6336 - val_accuracy: 0.8229 Epoch 19/50 28/28 [==============================] - 2s 77ms/step - loss: 0.6009 - accuracy: 0.8318 - val_loss: 1.6065 - val_accuracy: 0.5833 Epoch 20/50 28/28 [==============================] - 2s 76ms/step - loss: 0.6344 - accuracy: 0.7556 - val_loss: 0.6052 - val_accuracy: 0.8438 Epoch 21/50 28/28 [==============================] - 2s 76ms/step - loss: 0.1741 - accuracy: 0.9372 - val_loss: 0.4656 - val_accuracy: 0.8854 Epoch 22/50 28/28 [==============================] - 2s 76ms/step - loss: 0.1283 - accuracy: 0.9484 - val_loss: 0.4183 - val_accuracy: 0.8958 Epoch 23/50 28/28 [==============================] - 2s 78ms/step - loss: 0.1349 - accuracy: 0.9596 - val_loss: 0.8055 - val_accuracy: 0.8021 Epoch 24/50 28/28 [==============================] - 2s 76ms/step - loss: 0.1731 - accuracy: 0.9372 - val_loss: 0.8190 - val_accuracy: 0.8073 Epoch 25/50 28/28 [==============================] - 2s 76ms/step - loss: 0.1497 - accuracy: 0.9552 - val_loss: 0.5631 - val_accuracy: 0.8906 Epoch 26/50 28/28 [==============================] - 2s 78ms/step - loss: 0.1280 - accuracy: 0.9574 - val_loss: 0.7072 - val_accuracy: 0.8333 Epoch 27/50 28/28 [==============================] - 2s 76ms/step - loss: 0.2709 - accuracy: 0.9193 - val_loss: 0.3914 - val_accuracy: 0.9219 Epoch 28/50 28/28 [==============================] - 2s 77ms/step - loss: 0.1652 - accuracy: 0.9305 - val_loss: 0.5742 - val_accuracy: 0.8333 Epoch 29/50 28/28 [==============================] - 2s 77ms/step - loss: 0.0859 - accuracy: 0.9664 - val_loss: 0.4092 - val_accuracy: 0.9115 Epoch 30/50 28/28 [==============================] - 2s 77ms/step - loss: 0.0178 - accuracy: 0.9978 - val_loss: 0.4349 - val_accuracy: 0.9271 Epoch 31/50 28/28 [==============================] - 2s 75ms/step - loss: 0.0105 - accuracy: 1.0000 - val_loss: 0.4285 - val_accuracy: 0.9271 Epoch 32/50 28/28 [==============================] - 2s 77ms/step - loss: 0.0683 - accuracy: 0.9753 - val_loss: 0.6141 - val_accuracy: 0.8698 Epoch 33/50 28/28 [==============================] - 2s 78ms/step - loss: 0.2401 - accuracy: 0.9417 - val_loss: 0.7953 - val_accuracy: 0.8281 Epoch 34/50 28/28 [==============================] - 2s 76ms/step - loss: 0.2661 - accuracy: 0.9148 - val_loss: 0.8167 - val_accuracy: 0.8281 Epoch 35/50 28/28 [==============================] - 2s 76ms/step - loss: 0.0997 - accuracy: 0.9641 - val_loss: 0.5270 - val_accuracy: 0.9115 Epoch 36/50 28/28 [==============================] - 2s 76ms/step - loss: 0.0985 - accuracy: 0.9619 - val_loss: 0.4618 - val_accuracy: 0.9219 Epoch 37/50 28/28 [==============================] - 2s 75ms/step - loss: 0.0264 - accuracy: 0.9933 - val_loss: 0.4260 - val_accuracy: 0.9375 Epoch 38/50 28/28 [==============================] - 2s 77ms/step - loss: 0.0070 - accuracy: 1.0000 - val_loss: 0.4448 - val_accuracy: 0.9271 Epoch 39/50 28/28 [==============================] - 2s 77ms/step - loss: 0.0025 - accuracy: 1.0000 - val_loss: 0.4540 - val_accuracy: 0.9323 Epoch 40/50 28/28 [==============================] - 2s 77ms/step - loss: 0.0022 - accuracy: 1.0000 - val_loss: 0.4548 - val_accuracy: 0.9323 Epoch 41/50 28/28 [==============================] - 2s 77ms/step - loss: 0.0014 - accuracy: 1.0000 - val_loss: 0.4631 - val_accuracy: 0.9323 Epoch 42/50 28/28 [==============================] - 2s 83ms/step - loss: 0.0012 - accuracy: 1.0000 - val_loss: 0.4669 - val_accuracy: 0.9323 Epoch 43/50 28/28 [==============================] - 2s 78ms/step - loss: 0.0011 - accuracy: 1.0000 - val_loss: 0.4708 - val_accuracy: 0.9323 Epoch 44/50 28/28 [==============================] - 2s 76ms/step - loss: 0.0010 - accuracy: 1.0000 - val_loss: 0.4654 - val_accuracy: 0.9323 Epoch 45/50 28/28 [==============================] - 2s 77ms/step - loss: 9.7332e-04 - accuracy: 1.0000 - val_loss: 0.4633 - val_accuracy: 0.9323 Epoch 46/50 28/28 [==============================] - 2s 77ms/step - loss: 8.7667e-04 - accuracy: 1.0000 - val_loss: 0.4779 - val_accuracy: 0.9323 Epoch 47/50 28/28 [==============================] - 2s 76ms/step - loss: 8.4377e-04 - accuracy: 1.0000 - val_loss: 0.4851 - val_accuracy: 0.9323 Epoch 48/50 28/28 [==============================] - 2s 76ms/step - loss: 8.1836e-04 - accuracy: 1.0000 - val_loss: 0.4809 - val_accuracy: 0.9323 Epoch 49/50 28/28 [==============================] - 2s 76ms/step - loss: 7.3665e-04 - accuracy: 1.0000 - val_loss: 0.4744 - val_accuracy: 0.9323 Epoch 50/50 28/28 [==============================] - 2s 75ms/step - loss: 7.0643e-04 - accuracy: 1.0000 - val_loss: 0.4923 - val_accuracy: 0.9271
In [14]:
plt.figure(figsize=(10,7))
plt.subplot(1,2,1)
plt.plot(hist.history['accuracy'],label='accuracy')
plt.plot(hist.history['loss'],label='loss')
plt.legend()
plt.title("training set")
plt.grid()
plt.subplot(1,2,2)
plt.plot(hist.history['val_accuracy'],label='val_accuracy')
plt.plot(hist.history['val_loss'],label='val_loss')
plt.legend()
plt.title("validation set")
plt.grid()
plt.ylim((0,4))
Out[14]:
(0.0, 4.0)
Test set accuracy¶
In [15]:
score = vggmodel.evaluate(dx_test,y_test)
print("accuracy: ", score[1])
5/5 [==============================] - 0s 75ms/step - loss: 0.2725 - accuracy: 0.9375 accuracy: 0.9375
In [16]:
pred = vggmodel.predict(dx_test)
prediction = np.argmax(pred,axis=1)
true = np.argmax(y_test,axis=1)
best_prob = [pred[num,:][i] for num,i in enumerate(prediction)]
In [17]:
plt.figure(figsize = (9,8))
class_label = ['jute','maize','rice','sugarcane','wheat']
fig = sns.heatmap(confusion_matrix(true,prediction),cmap= "coolwarm",annot=True,vmin=0,cbar = False,
center = True,xticklabels=class_label,yticklabels=class_label)
fig.set_xlabel("Prediction",fontsize=30)
fig.xaxis.set_label_position('top')
fig.set_ylabel("True",fontsize=30)
fig.xaxis.tick_top()
In [18]:
def deepmodelpipeline(imagepath,model = vggmodel,label=[-1]):
pdict = {0:"jute",1:"maize",2:"rice",3:"sugarcane",4:"wheat"}
pred_x = deep_pipeline([imagepath])
prediction = model.predict(pred_x)
pred = np.argmax(prediction[0])
plt.imshow(plt.imread(imagepath))
if (label[0]!=-1):
plt.title("prediction : {0} % {1:.2f} \ntrue : {2}".format(pdict[pred],prediction[0,pred]*100,pdict[np.argmax(label)]))
else:
plt.title("prediction : {0}, % {1:.2f}".format(pdict[pred],prediction[0,pred]*100))
In [19]:
deepmodelpipeline('/kaggle/input/agriculture-crop-images/kag2/rice/rice024ahs.jpeg')
In [20]:
deepmodelpipeline('../input/agriculture-crop-images/kag2/wheat/wheat0004a.jpeg')
In [21]:
deepmodelpipeline("../input/agriculture-crop-images/kag2/maize/maize008ahf.jpeg")
In [22]:
deepmodelpipeline("../input/agriculture-crop-images/kag2/jute/jute005a.jpeg")
Prediction on test images¶
In [23]:
plt.figure(figsize=(20,20))
for num,path in enumerate(X_test[0:20]):
plt.subplot(4,5,num+1)
deepmodelpipeline(path,vggmodel,y_test[num])
In [24]:
plt.figure(figsize=(20,20))
for num,path in enumerate(X_test[20:40]):
plt.subplot(4,5,num+1)
deepmodelpipeline(path,vggmodel,y_test[num+20])
In [25]:
plt.figure(figsize=(20,20))
for num,path in enumerate(X_test[40:60]):
plt.subplot(4,5,num+1)
deepmodelpipeline(path,vggmodel,y_test[num+40])
In [26]:
plt.figure(figsize=(20,20))
for num,path in enumerate(X_test[100:120]):
plt.subplot(4,5,num+1)
deepmodelpipeline(path,vggmodel,y_test[num+100])
In [27]:
plt.figure(figsize=(20,20))
for num,path in enumerate(X_test[140:160]):
plt.subplot(4,5,num+1)
deepmodelpipeline(path,vggmodel,y_test[num+140])
In [28]:
vggmodel.save_weights("vggmodelweight.h5")
In [29]:
def resize_image(image_array):
return cv2.resize(image_array,(224,224))
def rescale_image(image_array):
return image_array*1./255
def read_image(image_path):
return plt.imread(image_path)
def plot_image(image_array):
try:
plt.imshow(image_array)
except:
plt.imshow(image_array[0])
def preprocess_image(image_path,reshape = True):
image = read_image(image_path)
image = resize_image(image)
image = rescale_image(image)
if(reshape == True):
image = image.reshape(-1,image.shape[0],image.shape[1],image.shape[2])
return image
def preprocess_imageslist(image_list):
imagelist = np.array([preprocess_image(img,reshape=False) for img in image_list])
return imagelist
def predict_and_plot(image,model):
pred_dict = {0:"jute",1:"maize",2:"rice",3:"sugarcane",4:"wheat"}
plt.imshow(image[0])
prediction = model.predict(image)
pred = pred_dict[np.argmax(prediction)]
plt.title(pred)
%matplotlib inline
def predict_and_plot5(imagelist,model):
pred_dict = {0:"jute",1:"maize",2:"rice",3:"sugarcane",4:"wheat"}
plt.figure(figsize=(20,10))
for num,image in enumerate(imagelist):
plt.subplot(1,5,num+1)
plt.imshow(image)
prediction = model.predict(image.reshape(-1,224,224,3))
pred = pred_dict[np.argmax(prediction)]
plt.title(pred)
def predict_and_plot51(imagelist,model):
pred_dict = {0:"jute",1:"maize",2:"rice",3:"sugarcane",4:"wheat"}
plt.figure(figsize=(100,10))
for i in range(0,55,5):
predict_and_plot5(imagelist[i:i+5],vggmodel)
def predict_and_plot30(imagelist,model):
plt.figure(figsize=(40,10))
for i in range(0,len(imagelist),5):
predict_and_plot5(imagelist[i:i+5],vggmodel)
In [30]:
testdata = pd.read_csv('../input/testssss/testdata.csv')
In [31]:
imagepath = testdata['testpath'].values
true_label = testdata['croplabel'].values
testdata.head()
Out[31]:
| Unnamed: 0 | testpath | crop | croplabel | |
|---|---|---|---|---|
| 0 | 0 | /kaggle/input/testssss/test_crop_image/wheat-f... | wheat | 4 |
| 1 | 1 | /kaggle/input/testssss/test_crop_image/jute-fi... | jute | 0 |
| 2 | 2 | /kaggle/input/testssss/test_crop_image/wheat-c... | wheat | 4 |
| 3 | 3 | /kaggle/input/testssss/test_crop_image/maize00... | maize | 1 |
| 4 | 4 | /kaggle/input/testssss/test_crop_image/rice-fi... | rice | 2 |
In [32]:
process_image = preprocess_imageslist(imagepath)
In [33]:
predict_and_plot51(process_image,model=vggmodel)
<Figure size 7200x720 with 0 Axes>
In [34]:
test_prediction = np.argmax(vggmodel.predict(process_image),axis=1)
In [35]:
accuracy_score(true_label,test_prediction)
Out[35]:
0.6666666666666666
In [36]:
plt.figure(figsize = (9,8))
class_label = ['jute','maize','rice','sugarcane','wheat']
fig = sns.heatmap(confusion_matrix(true_label,test_prediction),cmap= "coolwarm",annot=True,vmin=0,cbar = False,
center = True,xticklabels=class_label,yticklabels=class_label)
fig.set_xlabel("Prediction",fontsize=30)
fig.xaxis.set_label_position('top')
fig.set_ylabel("True",fontsize=30)
fig.xaxis.tick_top()
In [37]:
classify = true_label==test_prediction
wrong_classify = [num for num,p in enumerate(classify) if p==False]
correct_classify = [num for num,p in enumerate(classify) if p==True]
In [38]:
correct_class_image = process_image[correct_classify]
wrong_class_image = process_image[wrong_classify]
In [39]:
predict_and_plot30(wrong_class_image,vggmodel)
<Figure size 2880x720 with 0 Axes>
In [40]:
predict_and_plot30(correct_class_image,vggmodel)
<Figure size 2880x720 with 0 Axes>
In [41]:
labels = test_prediction.reshape(-1,1)
imagepath = imagepath.reshape(-1,1)
df = pd.DataFrame(np.append(imagepath,labels,axis=1),columns=['pathname','label'])
df.head()
Out[41]:
| pathname | label | |
|---|---|---|
| 0 | /kaggle/input/testssss/test_crop_image/wheat-f... | 4 |
| 1 | /kaggle/input/testssss/test_crop_image/jute-fi... | 0 |
| 2 | /kaggle/input/testssss/test_crop_image/wheat-c... | 4 |
| 3 | /kaggle/input/testssss/test_crop_image/maize00... | 1 |
| 4 | /kaggle/input/testssss/test_crop_image/rice-fi... | 2 |
In [42]:
df.to_csv('submission.csv')
In [43]:
get_image_path = "../input/testssss/test_crop_image/jute-field.jpg"
In [44]:
image1 = preprocess_image(get_image_path)
predict_and_plot(image1,model = vggmodel)
